You talk about extended RAM does not match the CMOS, but I can't see that you plugged in SIPPs into the sockets (between the last ISA slots)?!?

Maybe you can find the board somewhere here http://rk86.com/th99/m/m286_t.htm

If the sysinfo tools hang something is probably misconfigured, try changing the DIPs. There are not so many 😉

You're right, I confused the extended RAM with the area between 640KB and 1MB (which 286 systems seem to reserve for other stuff).

I couldn't find my board through TH99, but wherever I read about a four-SW block it seems to control color/monochrome (SW/1) and RAM size (SW2-4). Other options (RAM wait states, NPU sync/async, etc) are available by jumpers, but I haven't found my jumper setup yet...

Speaking of color/monochrome monitor, how exactly does that work? Whatever I choose I get color output (I'm using a TVGA8900C-based VGA card)...

Found something after all: http://stason.org/TULARC/pc/motherboards/A/AU … -BAM-12-G1.html (except for the CPU/chipset specs, the location of JP5 and an extra JP6 that I have somewhere near the FPU socket, it looks exactly like my board)

Norton Commander does start (using the /s or /t command line switches, for small or tiny memory start-up), but its System Information menu option will hang it. I found two more sysinfo tools that I could use (SYSCHK and SNOOPER), but getting them to that hard-drive using a floppy disk was a nightmare - I've only managed to successfully access the floppy drive once or twice so far, all other attempts would leave the FDD led on and the system frozen. FDD type is correctly configured in the BIOS, I tested three ISA controllers, two cables and three floppy drives, and I tried with both 0 and 1 wait state settings... eventually I used some 386 board to have those tools transferred to the hard-drive, then ran them on the 286 - they work.

Next step would be to figure out why floppy access doesn't work and to get 2 or 4 MB of SIPP memory...

Try removing the 287. As there is a jumber for the fpu sync/async mode, in order for it to run at a different speed than the cpu speed, the empty space next to the psu connector should be occupied by a 20MHz crystal for a 10MHz fpu. I used to experience problems like yours once on two different 286 boards using a cyrix 287xl. If that doesn't work, you should check the memory chips. Perhaps run memtest86.
Oh, and you'd better remove that battery.

This board is weird at times... woke up this morning and went straight for removing the 287; after that the floppy disk access worked just fine. Then I thought I'd recreate the problem, so I put the FPU back - it took me 7 or 8 attempts to make the board POST again (with or without FTP and with the video card placed in different ISA slots). Finally I got it running in yesterday's configuration (with the FPU) - the FDD is still working. I also tested another AMD P80C287-10 that I had around, and it was a success too. So right now floppy access is working with a FPU installed - whatever the fix was, I hope it's permanent...

Speaking of the FPU jumper, what does "sync" mean with 286/287? I've read that it usually means 287 freq = 2 / 3 * 286 freq, since internally the CPU divides the external clock by 2 (32 / 2 = 16 in my case), while the FPU divides it by 3 (32 / 3 =~ 10.7 here). Also, what version of memtest86 works on a 286? The latest memtest86+ apparently doesn't...

Battery will go out eventually, after I'm convinced that this is a stable, worth keeping, board 😀.

Well if that works that way (2/3 cpu speed) that's really not a problem then. I said that because my latest 286 board has those two modes : NPU mode select synchronous with CPU clock / NPU mode select synchronous with oscillator installed at OSC1.

As for memtest86, I'm sorry, it's my fault that I suggested that before trying to run it on a 286...

I seem to have found an eight-70ns-DIP20 chip set (taken from a Trident VLB video card) that the board is happy with. Now, another question: should I decide to populate the four DIP16 parity chips, do those have to be 70ns as well? As I said, I have around 30 DIP16 chips, but they are all labelled "-12" (120ns, right?)...

120ns indeed sounds like too slow.

But if you are only using it for gaming, why bother.

Apparently I spoke too soon 🙁. Yesterday I decided to put the board into a case and perform a fresh MS-DOS 6.22 install on some 1GB hard-drive (previously "prepared" with the Seagate version of Ontrack Disk Manager).

If I boot using floppy #1 _after_ Ontrack is initialized (boot from HDD, then "press Space to boot from diskette"), the system would either hang or reset right before copying files.

If I boot straight from the FDD the install process goes on into the file copying phase, but complains that many files (on all three disks) cannot be read. After the (incomplete) DOS eventually installs I'm able to copy and expand any file from any of the disks.

I could go with that and manually "complete" the installation (or I could perform the installation on some 386 or 486 and then move the HDD to the 286), but I'm very intrigued by this behavior...

Or, I could just leave the 286 for later and build a 386, just to be able to try DESQview, Coherent, SVR4... 😀

I think that 1gb is too large for a 286 to handle. It may be a limitation of the controller also. I remember trying with a 2gb CF card on my 386, that the BIOS auto-detected correctly, and which I had partitioned before on another machine. There were many problems. At the beginning it was ok, but when the data size on the disk was larger than 500MB it wouldn't read it correctly.

Well, I tried different I/O controllers (that I successfully used with MS-DOS on drives larger than 1GB), then I tried jumper-toggling between 0 and 1 wait state, then I tried installing with "turbo off" (8 MHz instead of 16), then I tried re-formatting and re-creating the first setup floppy disk.

Every time the files that setup couldn't read were the same, so it hit me - I compared the list of those files with a simple DIR output, and it turned out that those files were exactly the compressed ones; for some still unknown reason they cannot be EXPANDed during installation, but they can be EXPANDed just fine after the "minimal" and incomplete OS is started from the HDD.

Does SETUP.EXE have any command line parameters/options? Maybe 640KB is not enough and there is some "low memory mode" flag, or maybe there is a way to make it more verbose or to log errors somewhere...

I also found the link below (so it seems that SETUP.EXE wasn't always expected to work), but I'm wondering if what I'm experiencing is "normal" for a 286 system with 640KB of RAM...

http://support.microsoft.com/kb/117245

I'm sure I've booted dos 6.22 on a 286, but I don't remember actually installing it. On mine I've installed version 5.0. The fact that I've got it stored in my basement right now prevents me from trying it. I've got here another ancient 286/12 board, with 1mb RAM in 36 chips. It doesn't work right now, but I suspect it has to do with the memory chips. If I get it to post, I promise to try dos 6.22 and inform you.

I've been working on my 286 board, and found one memory chip that is definately faulty. There may be several others as it posts but it hangs when it is loading dos, so the only way it can work ok is when the memory is set to 640kb. Using my old conner 41MB HDD, I was able to install dos 6.22 fine.